The following example deployment shows two data centers each having its own DvR domain, connected through a backbone.
All nodes in data center Campus 1 belong to DvR domain shown in green, and the nodes in the data center Campus 3 belong to the DvR domain shown in orange. The two DvR domains are individually managed, so in this scenario, the controllers colored orange manage the orange Leaf nodes and the controllers colored green manage the green Leaf nodes. However, subnets can still be stretched across the DvR domains (and possibly between buildings), as shown in the figure.
Each DvR domain learns its own Layer 3 data and distributes this information to its own Leaf nodes. Layer 3 host information that is redistributed from other DvR Domains is learned by the Controllers only (through inter-DvR domain redistribution) and is programmed on the Leaf nodes in the same domain, but not in the other Domain. For example, Layer 3 information redistributed from domain 2 is learned by all controllers including the domain 1 controllers, but this information is not distributed to the Leaf nodes in domain 1.Hosts in one DvR domain can reach the hosts in the other DvR domain only through the Controllers.
All controllers in all domains are always part of the DvR backbone by default, as they are connected by the SPB Fabric. The DvR backbone connects many DvR domains.
Thus DvR can scale to multiple campuses, allowing a simplified way to deploy a large scale fully-routed infrastructure.